Phosphatidylinositol 4,5-bisphosphate regulates inspiratory burst activity in the neonatal mouse preBötzinger complex

Abstract

Neurons of the preBötzinger complex (preBötC) form local excitatory networks and synchronously discharge bursts of action potentials during the inspiratory phase of respiratory network activity. Synaptic input periodically evokes a Ca²⁺-activated non-specific cation current (I_CAN) postsynaptically to generate 10–30 mV transient depolarizations, dubbed inspiratory drive potentials, which underlie inspiratory bursts. The molecular identity of I_CAN and its regulation by intracellular signalling mechanisms during inspiratory drive potential generation remains unknown. Here we show that mRNAs coding for two members of the transient receptor potential (TRP) family of ion channels, namely TRPM4 and TRPM5, are expressed within the preBötC region of neonatal mice. Hypothesizing that the phosphoinositides maintaining TRPM4 and TRPM5 channel sensitivity to Ca²⁺ may similarly influence I_CAN and thus regulate inspiratory drive potentials, we manipulated intracellular phosphatidylinositol 4,5-bisphosphate (PIP₂) and measured its effect on preBötC neurons in the context of ongoing respiratory-related rhythms in slice preparations. Consistent with the involvement of TRPM4 and TRPM5, excess PIP₂ augmented the inspiratory drive potential and diminution of PIP₂ reduced it; sensitivity to flufenamic acid (FFA) suggested that these effects of PIP₂ were I_CAN mediated. Inositol 1,4,5-trisphosphate (IP₃), the product of PIP₂ hydrolysis, ordinarily causes IP₃ receptor-mediated I_CAN activation. Simultaneously increasing PIP₂ while blocking IP₃ receptors intracellularly counteracted the reduction in the inspiratory drive potential that normally resulted from IP₃ receptor blockade. We propose that PIP₂ protects I_CAN from rundown by interacting directly with underlying ion channels and preventing desensitization, which may enhance the robustness of respiratory rhythm.

Rhythmically active networks throughout the CNS generate activity patterns ranging from locomotor bursts (< 1 Hz) to gamma voltage oscillations (30–90 Hz) (Buzsáki, 2006). The characteristic features of these rhythms can often be attributed to specific ion channels and intracellular regulatory mechanisms that control the voltage trajectories of the constituent neurons and ultimately govern the form of the network output (Grillner, 2003, 2006; Bartos et al. 2007).

Breathing is a particularly important rhythmic behaviour in mammals that originates in the brainstem. Neurons within the ventral medullary site dubbed the preBötzinger complex (preBötC) synchronously discharge bursts of action potentials and drive inspiratory breathing movements (Smith et al. 1991; Rekling & Feldman, 1998; Feldman & Del Negro, 2006; Janczewski & Feldman, 2006). Although the rhythmogenic role of the preBötC is generally well accepted, the molecular mechanisms and intracellular signalling pathways that regulate inspiratory burst generation in preBötC neurons are not entirely clear.

Calcium-activated non-specific cation current (I_CAN) contributes postsynaptically to inspiratory burst generation (Pace et al. 2007a). The ion channel(s) underlying I_CAN may belong to the transient receptor potential (TRP) family of ion channels first identified in Drosophila (Minke, 1977). Members of the melastatin-like subfamily TRPM4 and TRPM5 are unique among TRP channels because they form flufenamic acid (FFA)-sensitive, Ca²⁺-impermeable monovalent cation channels gated by intracellular Ca²⁺ transients (Launay et al. 2002; Hofmann et al. 2003; Montell, 2005; Ullrich et al. 2005; Ramsey et al. 2006), which are properties expressed by I_CAN at the whole-cell level in a variety of cell types (Teulon, 2000) including preBötC neurons (Pace et al. 2007a). If TRPM4 and TRPM5 give rise to I_CAN in preBötC neurons, then their regulation could influence I_CAN on a cycle-to-cycle basis and thus shape inspiratory bursts during respiratory rhythm generation.

The function of TRPM4 and TRPM5 is intimately coupled to phosphoinositide signalling (Fig. 1). For example, phosphatidylinositol 4,5-bisphosphate (PIP₂) maintains the Ca²⁺ sensitivity required to activate TRPM4 and TRPM5 channels (Liu & Liman, 2003; Zhang et al. 2005; Nilius et al. 2006) and thus PIP₂ depletion leads to TRPM4 and TRPM5 desensitization (Liu & Liman, 2003; Nilius et al. 2006). Phosphoinositide (PI) signalling also evokes TRPM4 and TRPM5 channel currents (as well as I_CAN) because PIP₂ hydrolysis forms inositol 1,4,5-trisphosphate (IP₃) and causes IP₃ receptor-mediated intracellular Ca²⁺ release (Clapham, 2003; Hofmann et al. 2003), which is important for inspiratory bursts in preBötC neurons (Pace et al. 2007a). If TRPM4 and TRPM5 underlie I_CAN, then the cycling of PIP₂ and IP₃ could provide feedback mechanisms that regulate drive potential generation. PIP₂ hydrolysis should deplete PIP₂ stores and ultimately lead to TRPM4 and TRPM5 channel desensitization. Thus, determining how PI signalling regulates I_CAN may be important to understand how respiratory rhythm, which must function unceasingly to sustain breathing, remains robust and stable.

Figure 1. Model of I_CAN activation with hypothesized channel identity and PIP₂ regulation.

Glutamate (triangle) activates group 1 metabotropic glutamate receptors (mGluRs), which then cause G_q-proteins to stimulate phospholipase C (PLC). PLC hydrolyses PIP₂ into diacylgylcerol (DAG) and IP₃, which activates IP₃ receptors (IP₃R) causing Ca²⁺ release from the endoplasmic reticulum (ER). Intracellular Ca²⁺ gates TRPM4 and TRPM5 channels, causing Na⁺ influx and K⁺ efflux that results in depolarization from baseline membrane potentials of –60 mV. PIP₂ also acts directly on TRPM4 and TRPM5 channels to maintain Ca²⁺ sensitivity (dotted arrow).

Based on the likelihood that TRPM4 and TRPM5 channels give rise to I_CAN, here we test the two-part hypothesis that TRPM4 and TRPM5 channels are expressed within the preBötC region and that PIP₂ regulates inspiratory burst generation. By limiting I_CAN desensitization, we propose that PIP₂ promotes periodic recruitment of I_CAN and prevents the rundown of its inspiratory burst-generating function in preBötC neurons, which may help maintain robust breathing rhythms.

Methods

Slice preparation

We used neonatal C57BL/6 mice (postnatal day 1–7 (P1–7)) for in vitro electrophysiology as well as RT-PCR experiments. The Institutional Animal Care and Use Committee at The College of William and Mary approved all protocols.

Neonatal mice were anaesthetized by hypothermia and rapidly decerebrated prior to dissection in normal artificial cerebrospinal fluid (ACSF) containing (mm): 124 NaCl, 3 KCl, 1.5 CaCl₂, 1 MgSO₄, 25 NaHCO₃, 0.5 NaH₂PO₄ and 30 d-glucose, equilibrated with 95% O₂ and 5% CO₂ with pH = 7.4. Transverse slices (550 μm thick) containing the preBötC as well as hypoglossal (XII) motoneurons and premotoneurons were sectioned using a vibrating microslicer. The rostral surface was cut just above the XII nerve roots at the level of the dorsomedial cell column and principal lateral loop of the inferior olivary nucleus; the preBötC was at or near the rostral surface (Gray et al. 1999; Ruangkittisakul et al. 2006). The caudal cut captured the obex.

Electrophysiology

Slices were placed rostral surface up in a 0.5 ml recording chamber on a fixed-stage microscope equipped with Koehler illumination and perfused with 27°C ACSF at 4 ml min⁻¹. ACSF K⁺ concentration was raised to 9 mm and respiratory motor output was recorded from XII nerve roots using suction electrodes and a differential amplifier.

Differential interference contrast videomicroscopy (Inoue & Spring, 1997) was used to visualize neurons and control micropipette movements. Patch-clamp recordings were performed on neurons visualized in the preBötC region ventrally adjacent to the semi-compact division of the nucleus ambiguus (Gray et al. 1999; Wang et al. 2001; Ruangkittisakul et al. 2006). Inspiratory neurons discharge spikes superimposed upon a 200–500 ms envelope of depolarization dubbed the inspiratory drive potential that is coincident with XII discharge (Pace et al. 2007a,b). All preBötC neurons with robust inspiratory activity were suitable for experiments; no attempt to identify neurons with voltage-dependent pacemaker properties was made and expiratory neurons were excluded. Current-clamp recordings were performed using a Dagan IX2-700 amplifier (Minneapolis, MN, USA). Intracellular pipettes were fabricated from capillary glass (O.D., 1.5 mm; I.D., 0.87 mm) and filled with patch solution (contents listed below). Access resistance was compensated with bridge balance. Data were digitally acquired at 4–20 kHz using a 16-bit A/D converter (Powerlab, ADInstruments, Colorado Springs, CO, USA) after low-pass filtering at 1 kHz to avoid aliasing.

We continuously monitored membrane potential and adjusted the bias current to maintain a consistent baseline of –60 mV, which provides a uniform standard for comparing inspiratory drive potentials among preBötC neurons. Inspiratory bursts were digitally smoothed to remove spikes and facilitate measurements of the underlying inspiratory drive potential (see Fig. 1 in Pace et al. 2007b) using Chart v5.4 (AD Instruments). The Peak Parameters extension in Chart software measured inspiratory drive potential amplitude and area.

The standard potassium gluconate patch solution contained (mm): 140 potassium gluconate, 5 NaCl, 1 EGTA, 10 Hepes, 2 Mg-ATP (excluded in some trials, e.g. Figs 3B, D–F and 5B and C) and 0.3 Na-GTP (pH = 7.3 with KOH). Pipette resistance was 3–4 MΩ and a liquid junction potential of 8 mV was corrected offline. One (or sometimes two, e.g. Fig. 5B) drugs that affect PIP₂ or IP₃ signalling pathways were added to the patch solution and applied intracellularly at the following concentrations: 10 μm PIP_2–1,2-dicotanoyl-sn-glycerol (diC8-PIP₂) (Echelon Biosciences, Salt Lake City, UT, USA), 30 μg ml⁻¹ poly l-lysine (Sigma, St Louis, MO, USA), 50 μm wortmannin (Sigma), 1 μm Xestospongin-C (Sigma), or 10 μm PLC-resistant PIP₂ synthesized by H. Zhang and G. Prestwich (Zhang et al. 2006; analog 4).

Figure 3. The effects of excess PIP₂ on inspiratory drive potentials.

A, bar chart summarizing the effects of diC8-PIP₂ (diC8) on mean inspiratory drive potentials (±s.e.m.). Inspiratory drive potential amplitude and area are plotted as a per cent of control for the following conditions: 10 μm diC8 with Mg-ATP (dark grey bars) or diC8 without Mg-ATP (white bars). The same measure is shown for the effects of flufenamic acid (FFA, 100 μm) in the presence of diC8-PIP₂ (light grey bars). *Statistical significance at P < 0.05. B, representative traces showing the effects of Mg-ATP removal from the standard patch solution. Sequential traces represent perforated patch (pp), whole cell immediately after patch rupture (wc), and whole cell 30 min after patch rupture. Baseline membrane potential (V_m) was held at –60 mV. XII represents hypoglossal nerve motor output. Mg-ATP was excluded from all subsequent diC8 experiments in panels D–F. C, representative traces from a control experiment show drive potential stability and longevity under standard conditions in vitro. Standard patch solution was employed with Mg-ATP present. The drive potential did not change over 60 min. D, representative neuron showing diC8 enhancement of drive potential onset. Sequential traces show pp, wc and wc 5 min after diC8 dialysis. E, representative neuron showing diC8 prolongation of drive potentials and the effects of FFA. The sequence of traces are similar to D (above) with the additional FFA condition after 25 min of diC8 dialysis. Fa–c, the effects of diC8 and FFA on ectopic depolarizations in preBötC neurons. Fa, example of ectopic depolarizations observed immediately following patch rupture. Inset shows the superposition of ectopic depolarizations in control and diC8. Fb, the same cell 3 min after patch rupture. Fc, with diC8 still present, FFA was bath-applied and ectopic depolarizations disappeared entirely.

Figure 5. Excess PIP₂ augments drive potentials through an IP₃-independent mechanism.

A, intracellular Xestospongin (Xes, 1 μm) attenuates inspiratory drive potentials by blocking IP₃ receptors. Perforated-patch (pp) and whole-cell (wc) control conditions are illustrated along with whole-cell Xes dialysis at steady state. B, typical data from an inspiratory neuron in response to co-application of Xes and diC8-PIP₂ (diC8). The inspiratory burst changed shape somewhat but neither amplitude nor area changed significantly. C, typical response to 10 μm intracellular PLC-resistant PIP₂. Again, neither the amplitude nor the area of the inspiratory burst changed significantly. D, bar chart plots the mean amplitude of inspiratory drive potentials (±s.e.m.) as per cent of control in the following conditions: diC8 alone (left white bar), Xes alone (right white bar), co-applied diC8 and Xes (dark grey bar), and PLC-resistant PIP₂ (light grey bar). Here ** denotes P < 0.01 and ns denotes ‘not significant’ at P ≫ 0.05.

Control measurements of inspiratory drive potentials were obtained using nystatin-perforated patches. Nystatin (250 μg ml⁻¹) was added to potassium gluconate patch solution immediately prior to use and discarded after 120 min. The amplitude and area of the underlying inspiratory drive potentials could be accurately measured after ∼20 min of exposure to nystatin, even though the high impedance of the perforated patch partially attenuated action potentials (Pace et al. 2007a). Subsequently, drugs that target PIP₂ signalling pathways were delivered intracellularly via patch rupture, which dialyses the cytosol in the whole-cell configuration. Patch rupture also lowered access resistance and enabled action potentials to measure full amplitude. FFA (Sigma) was often bath-applied after intracellularly applied drugs achieved steady state.

The mean inspiratory drive potential and XII motor output were computed by averaging 10 consecutive inspiratory bursts. All samples were tested for normality using the Anderson–Darling test. Student's paired t tests and Wilcoxon signed ranks tests were applied to determine significance for normal and non-normal distributions, respectively, with minimum significance set at 0.05 or less. Experiments in which XII motor output changed significantly during the course of intracellular drug application were discarded.

RT-PCR

For reverse transcriptase polymerase chain reaction (RT-PCR) experiments, total RNA was purified from flash-frozen kidney tissue using RNeasy (Qiagen, Valencia, CA, USA). The preBötC kernel was dissected from bilateral regions of the slice preparation and immediately flash-frozen in liquid nitrogen. Total RNA from these samples was extracted with the RNaqueous Micro kit (Ambion, Austin, TX, USA). Possible genomic DNA contaminants were removed by incubating all RNA samples at 37°C with DNase (from RNAqueous kit or Promega, Madison, WI, USA) and either RNasin (Promega) or 1× Dnase I buffer (provided with RNAqueous kit). Using the iSCRIPT cDNA Synthesis kit (Bio-Rad, Hercules, CA, USA), cDNA was reverse transcribed from 1 μg of total RNA. Following the manufacturer's instructions, samples were incubated for 5 min at 25°C, 60 min at 42°C, and 5 min at 85°C. For negative controls, no reverse transcriptase was added to the reaction.

PCR was carried out using standard conditions: 0.5 μm of each primer, 1.5 mm MgCl₂, 0.2 mm deoxynucleoside triphosphates (dNTPs) and 1.25 U Supertaq polymerase (Ambion), although Supertaq buffer concentration was doubled (from 1×) for TRPM4 reactions to optimize the reaction. This conventional protocol was the basis for amplification: 5 min 94°C ‘hot start’ to prevent mis-priming, 40 cycles of 30 s at 94°C, 1–2 min primer annealing, and 1–2 min elongation at 72°C, followed by 7 min at 72°C, and finally holding at 4°C. These reaction conditions were modified as follows: TRPM4 reactions involved 1 min annealing at 60°C and 1 min elongation, TRPM5 reactions involved 2 min annealing at 57°C and 2 min elongation, and GADPH reactions involved 1 min annealing at 60°C and 2 min elongation. In order to assess the quality and presence of the cDNA, all samples were tested for a ubiquitously expressed gene, GADPH (accession no. NM_008084), using the following primers: Forward, 5′-ACCACAGTCCATGCCATCAC-3′; reverse, 5′-TCCACCACCCTGTTGCTGTA-3′ (Kunert-Keil et al. 2006). To test preBötC samples for the presence of TRPM4 (accession no. NM_175130) and TRPM5 (accession no. NM_020277), the primer sets: forward, 5′-GGCCCAAGATTGTCATAGTG-3′; reverse, 5′-TTGGCATACTGGGACACACA-3′ (Guinamard et al. 2004a; note: forward and reverse primers were switched in this report); and forward, 5′-TCCTGTTCATTG-TGGGAGTCAC-3′; reverse, 5′-TGGCGATCAGAAGG-TTCATG-3′ (Paulsen et al. 2000), were used, respectively. All primers were designed to cross at least one intron as an additional control to assess the unintended amplification of residual genomic DNA; amplification of genomic DNA would result in a recognizably higher molecular weight fragment. Fragments amplified by each set of primers were verified via direct sequencing of PCR products using the ABI BigDye 3.1 kit (Applied Biosystems, Foster City, CA, USA) and ABI 3100-Avant automated fluorescent sequencer.

Results

TRPM4 and TRPM5 expression

We used RT-PCR to detect the presence of TRPM4 and TRPM5 mRNA in the preBotC. Murine kidney tissue expressing TRPM4 and TRPM5 mRNA (Enklaar et al. 2000; Nilius et al. 2003; Kunert-Keil et al. 2006) served as positive controls for both genes.

We dissected the preBötC region bilaterally from slice preparations. Following total RNA extraction, all preBötC samples (n = 4) tested positive for the presence of TRPM4, TRPM5 and GADPH mRNA, whereas mRNA for these genes was absent in negative controls (Fig. 2). Sequencing and comparison to the National Center for Biotechnology Information (NCBI) database confirmed the identity of all RT-PCR products, which indicates that cells in the preBötC region express mRNA for TRPM4 and TRPM5 ion channels. These data are consistent with the proposal that TRPM4 and/or TRPM5 give rise to I_CAN in preBötC neurons (Pace et al. 2007a).

Figure 2. The mRNA coding for TRPM4 and TRPM5 is expressed in the preBötC region.

Total RNA was extracted from preBötC and positive control tissues and reverse transcribed. Amplified products of the expected sizes were obtained for TRPM4 (301 bp), TRPM5 (483 bp) and GADPH (452 bp). Negative control reactions were performed without reverse transcriptase and amplified nothing.

It is important to note that this technique does not precisely quantify mRNA levels, but rather reliably detects only its presence or absence. For example, in many cases, taking a broad region of kidney tissue as a positive control diluted TRPM5 mRNA levels, so its band appeared weaker than the TRPM5 band for preBötC tissue (e.g. Fig. 2, compare TRPM5 preBötC to positive control). Intensity of bands should not be taken to indicate levels of expression, but rather the presence or absence of expression. Nevertheless, these data clearly demonstrate that TRPM4 and TRPM5 mRNA are present in the preBötC.

PIP₂ manipulation affects inspiratory burst generation in preBötC neurons

Synaptic excitation during the inspiratory phase involves metabotropic glutamate receptors (see Fig. 1) that stimulate IP₃ production and trigger intracellular Ca²⁺ release to evoke I_CAN (Pace et al. 2007a). In addition to its role as a precursor to IP₃, we posit that PIP₂ prevents I_CAN desensitization by regulating its Ca²⁺ sensitivity, because PIP₂ has this effect on TRPM4 and TRPM5 ion channels (Liu & Liman, 2003; Zhang et al. 2005; Nilius et al. 2006). Therefore, we predicted that augmenting the supply of PIP₂ intracellularly would promote I_CAN activation via increasing IP₃ production and simultaneously preventing desensitization; the net effect would be to enhance inspiratory drive potentials.

We increased the concentration of intracellular PIP₂ in a single inspiratory neuron using patch solution containing 10 μm of the water-soluble PIP₂ analogue diC8-PIP₂. The area and amplitude of the inspiratory drive potentials measured in control via perforated patch matched those observed during the first few cycles of whole-cell recording before diC8-PIP₂ dialysis occurred (e.g. Fig. 3D and E). Therefore, control inspiratory drive potentials may be measured either during perforated-patch conditions (with attenuated spikes) or from the first few cycles in whole cell (with spikes full amplitude). After 10 min of diC8-PIP₂ dialysis, inspiratory drive potential amplitude and area increased to 126 ± 10% and 203 ± 16% of control, respectively (n = 3).

Mg-ATP restores Ca²⁺ sensitivity to TRPM4 in excised patches (Nilius et al. 2005), in much the same way as PIP₂ (Nilius et al. 2006), and thus we could not differentiate whether diC8-PIP₂ or Mg-ATP caused drive potential enhancement. Therefore, we removed Mg-ATP from the patch solution to isolate the putative effects of PIP₂ on I_CAN desensitization during inspiratory bursts and repeated the experiment.

As a control we removed Mg-ATP from the patch solution without adding diC8-PIP₂, which caused significant rundown in the inspiratory drive potential: amplitude and area decreased to 66 ± 12% and 58 ± 9% of control, respectively (P < 0.05, n = 8, Fig. 3B). In contrast, inspiratory drive potentials measured with potassium gluconate patch solution containing Mg-ATP did not change: amplitude and area were 100 ± 4% and 104 ± 6% of control, respectively (P ≫ 0.05, n = 6, Fig. 3C). These data show that Mg-ATP prevents drive potential rundown but does not, on its own, enhance inspiratory drive potentials.

Dialysing preBötC neurons with diC8-PIP₂ in the absence of Mg-ATP nevertheless augmented inspiratory drive potentials: the amplitude and area significantly increased to 126 ± 7% and 177 ± 15% of perforated-patch control (P < 0.05, n = 6, Fig. 3A, D and E). Therefore, augmentation conferred by diC8-PIP₂ is an independent effect unrelated to Mg-ATP.

Of the six preBötC neurons dialysed with diC8-PIP₂ in the absence of Mg-ATP, four neurons activated more rapidly and attained maximum amplitude earlier during the inspiratory phase compared with control (Fig. 3D) whereas two neurons extended inspiratory burst duration (Fig. 3E). Three of the six exhibited spontaneous ectopic subthreshold depolarizations in the interval between XII discharge within the first minute of whole-cell recording (0 min, Fig. 3Fa) that were greatly enhanced by diC8-PIP₂ after reaching steady state (= 3 min, Fig. 3Fb). This enhancement of the ectopic burst is emphasized in the inset in Fig. 3Fa, which overlays the control and diC8-PIP₂ responses for comparison.

If the effects of diC8-PIP₂ are attributable to I_CAN, then they should be sensitive to the I_CAN antagonist FFA. In steady-state diC8-PIP₂ conditions, bath application of FFA reduced inspiratory drive potentials to below control levels: the amplitude and area decreased to 78 ± 6% and 68 ± 7% of control, respectively (P < 0.05, n = 5, e.g. Fig. 3A, E and Fc). These data are commensurate with the attenuation caused by bath-applied FFA in prior experiments that did not involve PIP₂ manipulations (65–70% of control (their Fig. 6), Pace et al. 2007a). The ability of FFA to reverse the burst-augmenting effects of diC8-PIP₂ indicates that I_CAN is the primary current affected by excess PIP₂.

Whereas excess PIP₂ enhanced inspiratory burst generation (Fig. 3), we next sought to test whether PIP₂ depletion would diminish the ability of preBötC neurons to generate inspiratory bursts (Fig. 4). We included Mg-ATP in the potassium gluconate patch solution so that any change in inspiratory drive potentials would be attributable to PIP₂ reduction, and not the removal of Mg-ATP (e.g. Fig. 3C).

Figure 4. The effects of PIP₂ removal on inspiratory drive potentials.

A, sequential traces show the response of a representative inspiratory neuron to intracellular 30 μg ml⁻¹ poly l-lysine (PLL) and FFA. Perforated patch (pp) and whole cell immediately following patch rupture (wc) represent controls. PLL reached steady state after 20 min of whole-cell dialysis. Then, FFA was applied; the illustrated trace was taken at 30 min. The bar graph plots the mean drive potential amplitude and area (±s.e.m.) as per cent of control in response to PLL applied alone (grey bars) and in combination with FFA (white bars). B, sequential traces show the effects of intracellular application of 50 μm wortmannin (WM) and FFA on a representative inspiratory neuron. The control trace is superimposed (in grey) on top of the WM trace for comparison. The bar graph plots the mean drive potential amplitude and area (±s.e.m.) as per cent of control after application of WM alone (grey bars) and co-application of WM and FFA (white bars). For both graphs, ** denote statistical significance at P < 0.01 compared with pp or wc control.

Poly l-lysine (PLL) acts as a scavenger that reduces the amount of free PIP₂ in the plasma membrane (Suh & Hille, 2005) and attenuates TRPM4 currents in excised patches (Zhang et al. 2005; Nilius et al. 2006). Applied to preBötC neurons intracellularly with Mg-ATP present, 30 μg ml⁻¹ PLL significantly decreased inspiratory drive potential amplitude and area to 79 ± 6% and 70 ± 7% of perforated-patch control (amplitude P < 0.05, area P < 0.01, n = 7); a representative experiment is shown in Fig. 4A. Subsequent FFA application significantly reduced the amplitude and area of inspiratory drive potentials to 65 ± 7% and 36 ± 7% of control (P < 0.01, n = 7, Fig. 4A). These data suggest that PIP₂ depletion diminishes PI-sensitive mechanisms that contribute to drive potential generation, including (but not limited to) I_CAN. PLL scavenging does not completely block I_CAN because the effects of FFA were not occluded.

As a complementary experiment, we depleted PIP₂ using wortmannin, an inhibitor of PI 4-kinase that hinders PIP₂ production by preventing the synthesis of its precursor phosphatidylinositol 4-monophosphate from PI (Nakanishi et al. 1995) and causes TRPM4 channel desensitization (Zhang et al. 2005; Nilius et al. 2006). Applied intracellularly with Mg-ATP in the patch solution and a perforated-patch control, 50 μm wortmannin significantly decreased inspiratory drive potential amplitude and area to 65 ± 4% and 61 ± 6% of control, respectively (P < 0.01, n = 8, Fig. 4B). FFA application further attenuated inspiratory drive potential amplitude and area to 56 ± 6% and 38 ± 11% of control (P < 0.01, n = 7, Fig. 4B). These data suggest that inhibiting the synthesis of PIP₂ reduces the inspiratory burst-generating capabilities of preBötC neurons. However, wortmannin does not completely block the FFA-sensitive intrinsic current consistent with I_CAN, similar to an analogous experiment with PLL.

We showed that excess PIP₂ augments inspiratory drive potentials and that diminution of PIP₂ attenuates them. However, these effects may be explained in several ways. For example, PIP₂ may affect I_CAN by promoting IP₃-mediated Ca²⁺ release and I_CAN activation, or by interacting directly with the underlying channels to affect sensitivity to Ca²⁺, or through some combination of both pathways (Fig. 1). Additionally, excess PIP₂ may increase diacylgylcerol (DAG) levels, which regulates TRPM4 in cardiac myocytes (Guinamard et al. 2004a,b). Therefore, we designed experiments to differentiate the PIP₂-mediated effects on I_CAN desensitization from IP₃- and DAG-dependent I_CAN activation.

IP₃ receptor-mediated I_CAN activation is prevented by intracellularly applying the IP₃ receptor antagonist Xestospongin-C (Xes; 1 μm) (Gafni et al. 1997). In a prior study we reported that Xes reduced inspiratory drive potential amplitude and area to 67 ± 5% and 64 ± 6% of perforated-patch control (P < 0.01, n = 16) by acting to prevent IP₃-mediated Ca²⁺ release involved in I_CAN activation (Pace et al. 2007a). Original data are shown in Fig. 5A.

To test directly the role of PIP₂ in I_CAN desensitization, we coupled intracellular dialysis of diC8-PIP₂ with Xes (in the absence of Mg-ATP) to prevent IP₃ receptor-mediated Ca²⁺ release. If exogenous PIP₂ acts exclusively through the IP₃ pathway, and not by preventing I_CAN desensitization, then Xes should still attenuate inspiratory drive potentials, even in the presence of excess PIP₂ (see Fig. 5D (white bars) and Pace et al. 2007a). However, the combined dialysis of Xes and diC8-PIP₂ caused no statistically significant change compared with perforated-patch control (Fig. 5B): inspiratory drive potential amplitude and area measured 110 ± 8% and 130 ± 14% of control, respectively (P >> 0.05, n = 10, Fig. 5D, dark grey bars). These results suggest that the effects of excess PIP₂ (i.e. to augment inspiratory drive potentials as shown in Fig. 3A and D–F) are not solely attributable to increased IP₃ levels.

The protocol in Fig. 5B tests the direct effects of PIP₂ on I_CAN desensitization but does not preclude DAG signalling since Xes only blocks IP₃ receptors. Our subsequent test (Fig. 5C) precludes both IP₃ and DAG production. We intracellularly dialysed preBötC neurons with a water-soluble PIP₂ analogue that lacks the scissile P–O bond normally cleaved by PLC and thus cannot be hydrolysed into IP₃ and DAG. This PLC-resistant PIP₂ at high intracellular concentrations functions normally in maintaining Ca²⁺ sensitivity of TRPM4 (Zhang et al. 2006), and is predicted to outcompete with endogenous PIP₂ for PLC binding sites (W. Huang and G. D. Prestwich, unpublished molecular simulations) and thus occlude IP₃-dependent I_CAN activation by preventing IP₃ synthesis. Finally, if DAG does not contribute to the diC8-PIP₂ effects, then we predict commensurate results between PLC-resistant PIP₂ and co-application of Xes with diC8-PIP₂ (e.g. Fig. 5B).

The PLC-resistant PIP₂ (10 μm) caused no significant change in the inspiratory drive potential amplitude or area compared with perforated-patch control; these measures remained at 95 ± 8% and 97 ± 9% of control, respectively (P ≫ 0.05, n = 8). Figure 5C shows a representative experiment; mean data for amplitude are shown in part D. These data are consistent with PIP₂ directly influencing the Ca²⁺ sensitivity of I_CAN.

Discussion

Here we demonstrate that PIP₂ directly regulates inspiratory drive potentials during endogenous respiratory-related rhythmic activity. We propose that PIP₂ prevents long-term desensitization of I_CAN through interactions with underlying TRPM4 and TRPM5 channels.

Experimental approach to study I_CAN regulation in the context of inspiratory bursts

PIP₂ regulates TRPM4 and TRPM5 ion channels that give rise to I_CAN-like whole-cell currents (Liu & Liman, 2003; Zhang et al. 2005; Nilius et al. 2006). We now show that manipulating PIP₂ can both enhance as well as diminish inspiratory drive potentials in the context of endogenous respiratory-related behaviour in vitro.

Our objective is to understand the mechanisms that regulate inspiratory bursts, so we cannot study I_CAN using excised patch or whole-cell experiments from cell lines expressing only TRPM4 or TRPM5 ion channels. These experimental approaches are divorced from respiratory function in vitro. In addition, excised patch (Liu & Liman, 2003; Zhang et al. 2005; Nilius et al. 2006) and Ca²⁺ uncaging experiments (Ullrich et al. 2005) employ Ca²⁺ concentrations of 1 μm to 10 mm to evoke TRPM4 and TRPM5 currents, which may exaggerate Ca²⁺-dependent rundown of channel activity (Liu & Liman, 2003; Nilius et al. 2004). We do not yet know how these experimentally induced Ca²⁺ levels compare to the physiologically relevant Ca²⁺ fluctuations in preBötC neurons. Therefore we aimed to study PI regulation of inspiratory drive potentials, and by inference the role of I_CAN desensitization, with the endogenous preBötC rhythm intact.

I_CAN is the major charge carrier of the drive potential (Pace et al. 2007a), but it is not the only contributor pre- or postsynaptically. We used 100 μm FFA to ensure that our PIP₂ manipulations modified drive potentials via effects on I_CAN and not other intrinsic currents that contribute to drive potentials. FFA (100 μm) significantly attenuates I_CAN (Pace et al. 2007a). In various other neurons, this dose of FFA also affects: (i) gap junctions, (ii) NMDA receptors, (iii) Ca²⁺-activated Cl⁻ channels, and (iv) Ca²⁺-dependent K⁺ currents (I_K-Ca) (Ottolia & Toro, 1994; Greenwood & Large, 1995; Teulon, 2000; Harks et al. 2001; Srinivas & Spray, 2003; Wang et al. 2006). However, in preBötC neurons, 100 μm FFA had no effect on inspiratory drive potentials following intracellular blockade of I_CAN accomplished with the rapid intracellular Ca²⁺ buffer BAPTA (Pace et al. 2007a). These results indicate that the first three FFA side-effects (listed above) do not significantly influence inspiratory drive potential properties, and thus do not confound our use of FFA to test for the involvement of I_CAN in the current study.

We do not yet know whether FFA affects I_K-Ca in preBötC neurons (cf. Wang et al. 2006). Ca²⁺-dependent K⁺ currents contribute to burst termination in adult mammals, as shown in cats in vivo (Richter et al. 1993; Pierrefiche et al. 1995), but I_K-Ca does not significantly contribute to inspiratory drive potentials in preBötC neurons of neonatal mice (see Supplementary Fig. 1). Therefore, even if 100 μm FFA affects I_K-Ca, this side-effect would not be relevant to the periodic regulation of inspiratory drive potentials.

PIP₂ regulation of inspiratory drive potentials probably involves I_CAN

If excess PIP₂ augments drive potentials by enhancing I_CAN, then the I_CAN antagonist FFA should be equally effective if applied with diC8-PIP₂ as when applied alone. The combined application of diC8-PIP₂ and FFA decreased drive potentials to 65–75% of control; this is the same level of attenuation reported by Pace et al. (2007a) in experiments where FFA was applied alone. We conclude that excess intracellular PIP₂ enhances the burst-generating contribution of I_CAN.

Conversely, if PIP₂ removal acts principally to diminish I_CAN, then drive potential amplitude and area should decline in the presence of PLL or wortmannin, which we observed. One might predict that the inspiratory burst-generating role of I_CAN in preBötC neurons would be abolished entirely during PLL or wortmannin trials because the absence of PIP₂ has been shown to completely block TRPM4 and TRPM5 channel currents (Liu & Liman, 2003; Zhang et al. 2005; Nilius et al. 2006). However, the presence of Mg-ATP prevents complete TRPM4 channel desensitization after PIP₂ removal (Zhang et al. 2005), which is probably attributable to the necessary role of Mg-ATP in maintaining PI 4-kinases (Balla, 1998). Based on these findings, we expect a fraction of I_CAN in preBötC neurons to remain available following PIP₂ depletion and thus able to activate in response to intracellular Ca²⁺ attributable to NMDA receptor- or Ca²⁺ channel-mediated fluxes.

However, we are left with the conundrum that co-application of FFA with PLL or wortmannin decreased inspiratory drive potentials to a greater extent (area measured ∼36% of control, see Fig. 4) than FFA alone (∼65% of control, see Pace et al. 2007a). This suggests that PIP₂ removal might decrease the drive potential by affecting synaptic and intrinsic currents other than I_CAN, that are also sensitive to PIP₂ levels.

PIP₂ promotes activation of several ion channel types including HCN, TRPM7 and TRPM8 (Runnels et al. 2002; Rohacs et al. 2005; Suh & Hille, 2005; Zolles et al. 2006) so we cannot rule out the possibility that PIP₂ removal influences the burst-generating function of preBötC neurons via effects on multiple intrinsic membrane properties besides I_CAN. However, HCN channels that underlie the mixed cationic current I_h appear to influence respiratory frequency but not inspiratory burst generation (Mironov et al. 2000; Thoby-Brisson et al. 2000). It is currently unknown whether TRPM7 and/or TRPM8 channels are expressed in preBötC neurons and is beyond the scope of the present investigation. In addition, PIP₂ might be required for mGluR1-mediated mechanisms of inspiratory drive potential generation that involve K⁺ currents, not I_CAN (Pace et al. 2007a).

Furthermore, PLL and wortmannin undoubtedly cause rundown of intrinsic currents that depend on PKC phosphorylation because PIP₂ hydrolysis produces DAG, which in turn activates protein kinase C (PKC) (Hille, 2001). However, since the hydrolysable compound diC8-PIP₂ in the presence of Xes compared with the PLC-resistant PIP₂ analogue yielded similar results, we concluded that DAG is not a major factor in drive potential generation.

In addition to group I mGluRs, other important metabotropic receptors in preBötC neurons, including neurokinin 1 receptors (NK1Rs) and subtype 1 of the purinergic P2Y receptors (P2Y₁R) may activate the PLC pathway via G_q-protein and therefore utilize PIP₂ (Li et al. 1997; North & Barnard, 1997). However, the signalling pathways underlying NK1R and P2Y₁R actions in preBötC neurons remain to be determined; for example, NK1Rs can be coupled to a wide variety of G-proteins (Roush & Kwatra, 1998). Moreover, previous studies have shown that NK1Rs and P2Y₁Rs primarily regulate respiratory frequency (Gray et al. 1999; Pena & Ramirez, 2004; Lorier et al. 2007). In contrast, inspiratory drive potentials depend on metabotropic glutamate receptors coupled to PLC signalling, and this process is largely independent of frequency modulation (Pace et al. 2007a). Nevertheless, P2Y₁R and NK1R activation may both diminish available PIP₂ stores leading to less IP₃ production and I_CAN desensitization, which would attenuate the inspiratory drive potential. This is consistent with attenuations observed in the amplitude of preBötC field recordings during applications of agonists for both P2Y₁Rs and NK1Rs (Pena & Ramirez, 2004; Lorier et al. 2007).

TRPM4 and TRPM5 channels probably underlie I_CAN expressed by inspiratory neurons

We found mRNA coding for TRPM4 and TRPM5 channels in the preBötC region of the neonatal murine brainstem, despite previous reports that the brain is deficient in these genes (Perez et al. 2002; Nilius et al. 2003; Kunert-Keil et al. 2006). Our extraction methods provide a more sensitive test because we focused on a limited region of the ventral respiratory brain stem including the preBötC. Sampling the whole brain – even the whole brainstem – may dilute preBötC mRNA and thus increase the likelihood of a false negative.

Observing these genes in the preBötC is not sufficient to conclude that inspiratory neurons express functional TRPM4 and TRPM5 channels. Additionally, our protocol did not limit RNA extraction to preBötC inspiratory neurons but most probably sampled expiratory and non-respiratory neurons, as well as non-neural cells. It is also possible that post-transcriptional processing precludes the translation of one or both mRNAs. However, TRPM4 and TRPM5 are the only known channels that give rise to whole-cell currents with the biophysical properties of I_CAN present in preBötC neurons (Pace et al. 2007a) with respect to pharmacology, monovalent cation permeability and activation mechanism (Launay et al. 2002; Hofmann et al. 2003; Montell, 2005; Ullrich et al. 2005; Ramsey et al. 2006). While TRPM4 and TRPM5 differ in sensitivity to Ca²⁺ and FFA, fundamental properties of the two channels overlap substantially (Ullrich et al. 2005). We also documented that PIP₂ regulates I_CAN-dominated inspiratory drive potentials in a manner consistent with its effects on TRPM4 and TRPM5 channels. Therefore, we conclude that TRPM4 and TRPM5 channels are the ion channels underlying I_CAN in preBötC neurons.

Physiological role of PIP₂ in preBötC neurons

We propose that excess PIP₂ augments inspiratory drive potentials by counteracting I_CAN desensitization. Excess PIP₂ has no effect on non-desensitized TRPM5 channels (Liu & Liman, 2003) and pre-treatment with PIP₂ prevented TRPM4 desensitization from occurring (Nilius et al. 2006). Thus, PIP₂ counteracts desensitization but does not appear to augment other aspects of channel function. Our results with diC8-PIP₂ imply that I_CAN normally functions in a partially desensitized state in preBötC neurons during periodic respiratory activity.

TRPM4 and TRPM5 channels are mostly (or completely) deactivated at steady state when held at negative membrane potentials, which gives rise to their characteristic outwardly rectifying current–voltage curves (Hofmann et al. 2003; Liu & Liman, 2003; Nilius et al. 2003; Ullrich et al. 2005). PIP₂ reverses this desensitization for TRPM4 such that large inward currents gated by Ca²⁺ can flow at hyperpolarized membrane potentials (Nilius et al. 2006), and the same may be true for TRPM5 although the mechanism has not been fully documented (Liu & Liman, 2003). We propose that PIP₂ plays a similar role in preBötC neurons by alleviating (at least in part) I_CAN desensitization thus promoting its activation at negative voltages. This would enable I_CAN to participate to a greater extent in inspiratory burst generation, because inspiratory bursts initiate from baseline membrane potentials in the range –70 to –50 mV.

Consistent with this mechanism, diC8-PIP₂ frequently (4 cells out of 6 tested with diC8-PIP₂ and no Mg-ATP) augmented drive potentials at the onset of inspiratory bursts (see Fig. 3D). This suggests that bursts initiate at lower voltages (and possibly with smaller excitatory postsynaptic potentials as well) in the presence of excess PIP₂. Also supporting this interpretation, diC8-PIP₂ enabled small ectopic depolarizations to reach burst threshold (see Fig. 3Fa and b) and cause full-fledged inspiratory-like bursts. We conclude that the addition of exogenous PIP₂ augments inspiratory drive potential generation by lowering or removing the threshold of excitatory input necessary to activate I_CAN.

How might PIP₂-regulated I_CAN desensitization operate on a cycle-to-cycle basis to shape inspiratory bursts? If periodic PIP₂ depletion is involved in burst termination, then excess exogenous PIP₂ should stave off desensitization and delay inspiratory burst termination. However, this effect was only observed in 2 of the 6 neurons treated with diC8-PIP₂ (see Fig. 3E). We conclude that PIP₂ depletion (during IP₃ production) is not a major factor that governs inspiratory burst termination.

I_CAN in preBötC neurons normally functions in a partially desensitized state during rhythmic activity in vitro. Moreover, PIP₂ does not act on a cycle-to-cycle basis to cause burst termination nor promote burst onset. These two points are not in conflict; we theorize that PIP₂ maintains I_CAN functionality for the ‘long haul’, i.e. respiratory oscillations are meant to continue without lapse, conferring stability and longevity to inspiratory burst generation in a critical physiological system that necessitates both.